Jeuk, Sebastian; (2019) A tenant-aware identification scheme for Cloud Computing -- Universal Cloud Classification (UCC). Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Jeuk_000_Thesis_redacted_2.pdf - Accepted Version Download (14MB) | Preview

Abstract

Cloud has profoundly changed the provisioning and consumption of computing services across the Internet. It enables dynamic, scalable and on-demand deployment of services to allow users access to storage, compute and network resource capacities from anywhere. Cloud has adopted legacy technologies for traffic identification, such as VLAN and VxLAN, which were not originally designed for the cloud. These technologies come with known limitations, some of which are amplified by cloud characteristics. In this thesis, I analyse and compare entity and traffic flow identification technologies for general networks and the cloud. I investigate the deficiencies of legacy technologies used for traffic flow identification in the cloud and their incapability to cope with the emerging demands posed by cloud computing. My research reveals that the root cause of these problems is the lack of tenant-specific entity identification in the cloud. I argue that not a modification of an existing technology, but rather a new, overall solution is needed to address these challenges. I propose Universal Cloud Classification (UCC) as the next-generation tenant-specific entity identification scheme for cloud, which enables scalable, global, universal, consistent and unique cloud identification, including services, tenants as well as traffic flows within a cloud, between clouds and across the Internet. My research leads to a practical solution capable of assigning and managing the UCC identifiers both globally and locally. I test and evaluate the feasibility and performance of UCC in an environment using industrial standard hardware and software and realistic traffic scenarios. My results illustrate the practicality of UCC and highlight the superiority of UCC over existing technologies in terms of identification capabilities. I demonstrate that although VxLAN can practically identify only a few thousand tenants, UCC is able to identify millions of tenants, services and clouds, respectively. Even when we evaluate UCC and VxLAN for identifying the same number of tenants, UCC still either outperforms or shows similar performance than VxLAN in terms of line-rate, packets-per-second, jitter and end-toend delay. UCC is designed to support diverse implementations complimentary to the basic IPv4 and IPv6 options. Thanks to its technological advances, such as broad applicability, interoperability and implementation independency, UCC brings a wide range of new applications and benefits to cloud. It not only supports existing cloud structures, but is also flexible enough to accommodate new cloud deployment types, such as micro-service based applications. I also propose UCC as a Service (UCCaaS), which can help to drive timely and wide adoption of UCC across data centres. UCC has received significant attention from academia and industry. A series of patents have been filed and were granted through industry support with the aim to productise UCC for next-generation clouds.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item